Distant Starlight

The oldest proposal, and internally consistent — an omnipotent creator could produce photons mid-flight.

The consequence is specific. SN 1987A is 168,000 light years away, so on this view the light carrying the explosion was created in transit, and no star ever exploded. The neutrino burst detected hours before the visible light, the expanding debris observed since, and the cobalt-56 decay curve would all be part of a record of an event that did not occur.

The same applies to every variable star, every observed supernova, and every galaxy collision.

Answers in Genesis rejects this argument on theological grounds, on the basis that it makes God the author of a false record.

Barry Setterfield proposed in 1981 that c has decayed, and that historical measurements show a downward trend.

The historical trend is an artefact of improving instrumentation, and the apparent decline disappears when measurement uncertainties are accounted for.

The physical consequences are severe. The speed of light appears in E = mc², so a much larger c would change the energy output of stars by a corresponding factor, and in atomic energy levels, which would shift spectral lines. Spectra from distant galaxies show the same line positions as laboratory measurements, indicating the constants were the same when that light was emitted.

The fine structure constant, which combines c with other constants, has been measured in the Oklo natural nuclear reactor in Gabon — a 1.7-billion-year-old fission site — and found unchanged to within a very small margin.

Humphreys' white hole cosmology, the most technically developed proposal. It places Earth near the centre of a bounded distribution of matter, so that during a period of expansion, clocks on Earth ran slowly relative to distant clocks — billions of years passing in the distant universe while days passed on Earth.

Time dilation is real physics and the general approach is not absurd.

The specific model has been criticised by other creationists as well as by physicists. Samuel Conner and Hugh Ross identified errors in the treatment of the event horizon, and Humphreys revised the model in response.

The model also requires Earth to be near the centre of a bounded universe, which conflicts with the observed isotropy of the cosmic microwave background, and it does not clearly explain why events observed in distant galaxies appear to unfold at normal rates.

Jason Lisle's anisotropic synchrony convention, and it rests on a genuine feature of relativity. The one-way speed of light cannot be measured without already having synchronised clocks at two locations, so only the round-trip speed is empirically fixed. Choosing light to be infinitely fast toward the observer and c/2 away is a legitimate convention.

The physics is correct. Einstein discussed the conventionality of simultaneity, and Lisle is not misstating it.

The objection is that a convention cannot change what is observed. Adopting a different coordinate system relabels events without altering their physical relationships; if the observations are consistent under one convention with a 13.8-billion-year history, they remain so under another.

The convention also has to be applied consistently, and doing so produces an anisotropic universe in which physics differs by direction — which is not observed.

Parallax does not. It is trigonometry using a baseline of Earth's orbital diameter, which is measured by radar.

SN 1987A's light-echo distance uses only the geometry of the ring and the timing of its illumination.

The higher rungs of the ladder are calibrated against these geometric methods where they overlap, and the calibration is checked at each stage. Nothing in the chain requires assuming an age; the age is a conclusion from the distances.